The Zettelkasten Method

Early this year, Conor White-Sul­li­van in­tro­duced me to the Zet­telkas­ten method of note-tak­ing. I would say that this sig­nifi­cantly in­creased my re­search pro­duc­tivity. I’ve been say­ing “at least 2x”. Nat­u­rally, this sort of thing is difficult to quan­tify. The truth is, I think it may be more like 3x, es­pe­cially along the di­men­sion of “pro­duc­ing ideas” and also “early-stage de­vel­op­ment of ideas”. (What I mean by this will be­come clearer as I de­scribe how I think about re­search pro­duc­tivity more gen­er­ally.) How­ever, it is also very pos­si­ble that the method pro­duces se­ri­ous bi­ases in the types of ideas pro­duced/​de­vel­oped, which should be con­sid­ered. (This would be difficult to quan­tify at the best of times, but also, it should be noted that other fac­tors have dra­mat­i­cally de­creased my over­all re­search pro­duc­tivity. So, un­for­tu­nately, some­one look­ing in from out­side would not see an over­all boost. Still, my im­pres­sion is that it’s been very use­ful.)

I think there are some spe­cific rea­sons why Zet­telkas­ten has worked so well for me. I’ll try to make those clear, to help read­ers de­cide whether it would work for them. How­ever, I hon­estly didn’t think Zet­telkas­ten sounded like a good idea be­fore I tried it. It only took me about 30 min­utes of work­ing with the cards to de­cide that it was re­ally good. So, if you’re like me, this is a cheap ex­per­i­ment. I think a lot of peo­ple should ac­tu­ally try it to see how they like it, even if it sounds ter­rible.

My plan for this doc­u­ment is to first give a short sum­mary and then an overview of Zet­telkas­ten, so that read­ers know roughly what I’m talk­ing about, and can pos­si­bly ex­per­i­ment with it with­out read­ing any fur­ther. I’ll then launch into a longer dis­cus­sion of why it worked well for me, ex­plain­ing the spe­cific habits which I think con­tributed, in­clud­ing some de­scrip­tions of my pre­vi­ous ap­proaches to keep­ing re­search notes. I ex­pect some of this may be use­ful even if you don’t use Zet­telkas­ten—if Zet­telkas­ten isn’t for you, maybe these ideas will nonethe­less help you to think about op­ti­miz­ing your notes. How­ever, I put it here pri­mar­ily be­cause I think it will boost the chances of Zet­telkas­ten work­ing for you. It will give you a more con­crete pic­ture of how I use Zet­telkas­ten as a think­ing tool.

Very Short Summary


  • Sta­ples in­dex-cards-on-a-ring or equiv­a­lent, pos­si­bly with:

    • plas­tic rings rather than metal

    • differ­ent 3x5 in­dex cards (I recom­mend blank, but, other pat­terns may be good for you) as desired

    • some kind of divider

      • I use yel­low in­dex cards as di­viders, but slightly larger cards, tabbed cards, plas­tic di­viders, etc. might be better

    • qual­ity hole punch (if you’re us­ing differ­ent cards than the pre-punched ones)

  • qual­ity writ­ing in­stru­ment—must suit you, but,

    • multi-color click pen recommended

    • hi-tec-c co­leto es­pe­cially recommended


  • Num­ber pages with alphanu­meric strings, so that pages can be sorted hi­er­ar­chi­cally rather than lin­early -- 11a goes be­tween 11 and 12, 11a1 goes be­tween 11a and 11b, et cetera. This al­lows pages to be eas­ily in­serted be­tween other pages with­out mess­ing up the ex­ist­ing or­der­ing, which makes it much eas­ier to con­tinue top­ics.

  • Use the alphanu­meric page iden­ti­fiers to “hy­per­link” pages. This al­lows sub-top­ics and tan­gents to be eas­ily split off into new pages, and also al­lows for re­lated ideas to be in­ter­linked.

Be­fore I launch into the proper de­scrip­tion of Zet­telkas­ten, here are some other re­sources on note-tak­ing which I looked at be­fore div­ing into us­ing Zet­telkas­ten my­self. (Feel free to skip this part on a first read­ing.)

Re­lated Literature

There are other de­scrip­tions of Zet­telkas­ten out there. I mainly read How to Take Smart Notes, which is the best book on Zet­telkas­ten as far as I know—it claims to be the best write-up available in English, any­way. The book con­tains a thor­ough de­scrip­tion of the tech­nique, plus a lot of “philo­soph­i­cal” stuff which is in­tended to help you ap­proach it with the right mind­set to ac­tu­ally in­te­grate it into your think­ing in a use­ful way. I am sym­pa­thetic to this ap­proach, but some of the con­tent seems like bad sci­ence to me (such as the de­scrip­tion of growth mind­set, which didn’t strike me as at all ac­cu­rate—I’ve read some of the origi­nal re­search on growth mind­set).

An is­sue with some other write-ups is that they fo­cus on im­ple­ment­ing Zet­telkas­ten-like sys­tems digi­tally. In fact, Conor White-Sul­li­van, who I’ve already men­tioned, is work­ing on a Work­flowy/​Dy­nal­ist-like digi­tal tool for think­ing, in­spired par­tially by Zet­telkas­ten (and also by the idea that a Work­flowy/​Dy­nal­ist style tool which is de­signed ex­plic­itly to nudge users into good think­ing pat­terns with aware­ness of cog­ni­tive bi­ases, good prac­tices for ar­gu­ment map­ping, etc. could be very valuable). You can take a look at his tool, Roam, here. He also wrote up some thoughts about Zet­telkas­ten in Roam. How­ever, I strongly recom­mend try­ing out Zet­telkas­ten on ac­tual note-cards, even if you end up im­ple­ment­ing it on a com­puter. There’s some­thing good about it that I don’t fully un­der­stand. As such, I would ad­vise against trust­ing other peo­ple’s at­tempts to dis­till what makes Zet­telkas­ten good into a digi­tal for­mat—bet­ter to try it your­self, so that you can then judge whether al­ter­nate ver­sions are im­prove­ments for you. The ver­sion I will de­scribe here is fairly close to the origi­nal.

I don’t strongly recom­mend my own write-up over what’s said in How to Take Smart Notes, par­tic­u­larly the parts which de­scribe the ac­tual tech­nique. I’m writ­ing this up partly just so that there’s an eas­ily link­able doc­u­ment for peo­ple to read, and partly be­cause I have some ideas about how to make Zet­telkas­ten work for you (based on my own pre­vi­ous note-tak­ing sys­tems) which are differ­ent from the book.

Another source on note-tak­ing which I recom­mend highly is Lion Kim­bro’s How to Make a Com­plete Map of Every Thought You Think (html, pdf). This is about a com­pletely differ­ent sys­tem of note-tak­ing, with differ­ent goals. How­ever, it con­tains a wealth of in­spiring ideas about note-tak­ing sys­tems, in­clud­ing valuable tips for the raw phys­i­cal as­pects of keep­ing pa­per notes. I recom­mend read­ing this in­ter­view with Lion Kim­bro as a “teaser” for the book—he men­tions some things which he didn’t in the ac­tual book, and it serves some­what as “the miss­ing in­tro­duc­tion” to the book. (You can skip the part at the end about wikis if you don’t find it in­ter­est­ing; it is sort of out­dated spec­u­la­tion about the fu­ture of the web, and it doesn’t get back to talk­ing about the book.) Part of what I love about How to Make a Com­plete Map of Every Thought You Think is the manic brain-dump writ­ing style—it is a book which feels very “al­ive” to me. If you find its style grat­ing rather than en­gag­ing, it’s prob­a­bly not worth you read­ing through.

I should also men­tion an­other re­cent post about Zet­telkas­ten here on LW.

Zet­telkas­ten, Part 1: The Basics

Zet­telkas­ten is Ger­man for ‘slip-box’, IE, a box with slips of pa­per in it. You keep ev­ery­thing on a bunch of note cards. Nik­las Luh­mann de­vel­oped the sys­tem to take notes on his read­ing. He went on to be an in­cred­ibly pro­lific so­cial sci­en­tist. It is hard to know whether his pro­duc­tivity was tied to Zet­telkas­ten, but, oth­ers have re­ported large pro­duc­tivity boosts from the tech­nique as well.

Small Pie­ces of Paper Are Just Mo­du­lar Large Pie­ces of Paper

You may be think­ing: aren’t small pieces of pa­per bad? Aren’t large note­books just bet­ter? Won’t small pages make for small ideas?

What I find is that the drive for larger pa­per is bet­ter-served by split­ting things off into new note cards. Note-cards rele­vant to your cur­rent think­ing can be spread on a table to get the same big-pic­ture overview which you’d get from a large sheet of pa­per. Writ­ing on an ac­tual large sheet of pa­per locks things into place.

When I was learn­ing to write in my teens, it seemed to me that pa­per was a prison. Four walls, right? And the ideas were con­stantly try­ing to es­cape. What is a paren­the­sis but an idea try­ing to es­cape? What is a foot­note but an idea that tried—that jumped off the cliff? Be­cause pa­per en­forces sin­gle se­quence—and there’s no room for di­gres­sion—it im­poses a par­tic­u­lar kind of or­der in the very na­ture of the struc­ture.
-- Ted Nel­son, demon­stra­tion of Xanadu space

I use 3x5 in­dex cards. That’s quite small com­pared to most note­books. It may be that this is the right size for me only be­cause I already have very small hand­writ­ing. I be­lieve Luh­mann used larger cards. How­ever, I ex­pected it to be too small. In­stead, I found the small cards to be free­ing. I strongly recom­mend try­ing 3x5 cards be­fore try­ing with a larger size. In fact, even smaller sizes than this are vi­able—one early reader of this write-up de­cided to use half 3x5 cards, so that they’d fit in mtg deck boxes.

Writ­ing on small cards forces cer­tain habits which would be good even for larger pa­per, but which I didn’t con­sider un­til the small cards made them nec­es­sary. It forces ideas to be bro­ken up into sim­ple pieces, which helps to clar­ify them. Break­ing up ideas forces you to link them to­gether ex­plic­itly, rather than rely­ing on the lin­ear struc­ture of a note­book to link to­gether chains of thought.

Once you’re forced to adopt a link­ing sys­tem, it be­comes nat­u­ral to use it to “break out of the prison of the page”—tan­gents, par­en­thet­i­cals, ex­plana­tory re­marks, caveats, … ev­ery­thing be­comes a new card. This gives your thoughts much more “sur­face area” to ex­pand upon.

On a com­puter, this is es­sen­tially the wiki-style [[magic link]] which links to a page if the page ex­ists, or cre­ates the page if it doesn’t yet ex­ist—a crit­i­cal but all-too-rare fea­ture of note-tak­ing soft­ware. Again, though, I strongly recom­mend try­ing the sys­tem on pa­per be­fore jump­ing to a com­puter; putting your­self in a po­si­tion where you need to link in­for­ma­tion like crazy will help you to see the value of it.

This brings us to one of the defin­ing fea­tures of the Zet­telkas­ten method: the ad­dress­ing sys­tem, which is how links be­tween cards are es­tab­lished.

Paper Hypertext

We want to use card ad­dresses to or­ga­nize and refer­ence ev­ery­thing. So, when you start a new card, its ad­dress should be the first thing you write—you never want to have a card go with­out an ad­dress. Choose a con­sis­tent lo­ca­tion for the ad­dresses, such as the up­per right cor­ner. If you’re us­ing multi-color pens, like me, you might want to choose one color just for ad­dresses.

Wiki-style links tend to use the ti­tle of a page to refer­ence that page, which works very well on a com­puter. How­ever, for a pen-and-pa­per hy­per­text sys­tem, we want to op­ti­mize sev­eral things:

  • Easy lookup: we want to find refer­enced cards as eas­ily as pos­si­ble. This en­tails sort­ing the cards, so that you don’t have to go dig­ging; find­ing what you want is as easy as find­ing a word in the dic­tio­nary, or find­ing a page given the page num­ber.

  • Easy to sort: I don’t know about you, but for me, putting things in alpha­bet­i­cal or­der isn’t the eas­iest thing. I find my­self recit­ing the alpha­bet pretty of­ten. So, I don’t re­ally want to sort cards alpha­bet­i­cally by ti­tle.

  • Easy to write: an­other rea­son not to sort alpha­bet­i­cally by ti­tle is that you want to refer­ence cards re­ally eas­ily. You prob­a­bly don’t want to write out full ti­tles, un­less you can keep the ti­tles re­ally short.

  • Fixed ad­dresses: What­ever we use to refer­ence a card, it must re­main fixed. Other­wise, refer­ences could break when things change. No one likes bro­ken links!

  • Re­lated cards should be near each other. Alpha­bet­i­cal or­der might put closely re­lated cards very far apart, which gets to be cum­ber­some as the col­lec­tion of cards grows—even if look-up is quite con­ve­nient, it is nicer if the re­lated cards are already at hand with­out pur­pose­fully de­cid­ing to look them up.

  • No pre­set cat­e­gories. Creat­ing a sys­tem of cat­e­gories is a com­mon way to place re­lated con­tent to­gether, but, it is too hard to know how you will want to cat­e­go­rize ev­ery­thing ahead of time, and the needs of an ad­dress­ing sys­tem make it too difficult to change your cat­e­gory sys­tem later.

One sim­ple solu­tion is to num­ber the cards, and keep them in nu­mer­i­cal or­der. Num­bers are easy to sort and find, and are very com­pact, so that you don’t have the is­sue of writ­ing out long names. How­ever, al­though re­lated con­tent will be some­what nearby (due to the fact that we’re likely to cre­ate sev­eral cards on a topic at the same time), we can do bet­ter.

The essence of the Zet­telkas­ten ap­proach is the use of re­peated dec­i­mal points, as in “22.3.14”—cards ad­dressed 2.1, 2.2, 2.2.1 and so on are all thought of as “un­der­neath” the card num­bered 2, just as in the fa­mil­iar sub­sec­tion-num­ber­ing sys­tem found in many books and pa­pers. This al­lows us to in­sert cards any­where we want, rather than only at the end, which al­lows re­lated ideas to be placed near each other much more eas­ily. A card sit­ting “un­der­neath” an­other can loosely be thought of as a com­ment, or a con­ti­tu­a­tion, or an as­so­ci­ated thought.

How­ever, for the sake of com­pact­ness, Zet­telkas­ten ad­dresses are usu­ally writ­ten in an alphanu­meric for­mat, so that rather than writ­ing 1.1.1, we would write 1a1; rather than writ­ing 1.2.3, we write 1b3; and so on. This no­ta­tion al­lows us to avoid writ­ing so many pe­ri­ods, which grows tire­some.

Alter­nat­ing be­tween num­bers and let­ters in this way al­lows us to get to two-digit num­bers (and even two-digit let­ters, if we ex­haust the whole alpha­bet) with­out need­ing pe­ri­ods or dashes or any such sep­a­ra­tors to in­di­cate where one num­ber ends and the next be­gins.

Let’s say I’m writ­ing lin­early—some­thing which could go in a note­book. I might start with card 11, say. Then I pro­ceed to card 11a, 11b, 11c, 11d, etc. On each card, I make a note some­where about the pre­vi­ous and next cards in se­quence, so that later I know for sure how to fol­low the chain via ad­dresses.

Later, I might have a differ­ent branch-off thought from 11c. This be­comes 11c1. That’s the magic of the sys­tem, which you can’t ac­com­plish so eas­ily in a lin­ear note­book: you can just come back and add things. Th­ese tan­gents can grow to be larger than the origi­nal.

Don’t get too caught up in what ad­dress to give a card to put it near rele­vant ma­te­rial. A card can be put any­where in the ad­dress sys­tem. The point is to make things more con­ve­nient for you; noth­ing else mat­ters. Ideally, the tree would perfectly re­flect some kind of con­cep­tual hi­er­ar­chy; but in prac­tice, card 11c might turn out to be the pri­mary thing, with card 11 just serv­ing as a his­tor­i­cal record of what seeded the idea.

Similarly, a lin­ear chain of writ­ing doesn’t have to get a nice lin­ear chain of ad­dresses. I might have a train of thought which goes across cards 11, 11a, 11b, 11b1, 11b1a, 11b1a1, 18, 18a… (I write a lot of “1a1a1a1a”, and it is some­times bet­ter to jump up to a new top-level num­ber to keep the ad­dresses from get­ting longer.)

Mostly, though, I’ve writ­ten less and less in lin­ear chains, and more and more in branch­ing trees. Some­times a thought just nat­u­rally wants to come out lin­early. But, this tends to make it more difficult to re­view later—the cards aren’t split up into atomic ideas, in­stead flow­ing into each other.

If you don’t know where to put some­thing, make it a new top-level card. You can link it to what­ever you need via the ad­dress­ing sys­tem, so the cost of putting it in a sub­op­ti­mal lo­ca­tion isn’t worth wor­ry­ing about too much! You don’t want to be con­strained by the ideas you’ve had so far. Or, to put it a differ­ent way: it’s like start­ing a new page in a note­book. Zet­telkas­ten is sup­posed to be less re­stric­tive than a note­book, not more. Don’t get locked into place by try­ing to make the ad­dresses perfectly re­flect the log­i­cal or­ga­ni­za­tion.

Phys­i­cal Is­sues: Card Storage

Lin­ear notes can be kept in any kind of pa­per note­book. Non­lin­ear/​mod­u­lar sys­tems such as Zet­telkas­ten, on the other hand, re­quire some sort of binder-like sys­tem where you can in­sert pages at will. I’ve tried a lot of differ­ent things. Bin­ders are typ­i­cally just less com­fortable to write in (be­cause of the rings—this is an­other point where the fact that I’m left-handed is very sig­nifi­cant, and right-handed read­ers may have a differ­ent ex­pe­rience).

(One thing that’s im­proved my life is re­al­iz­ing that I can use a binder “back­wards” to get es­sen­tially the right-han­der’s ex­pe­rience—I write on the “back” of pages, start­ing from the “end”.)

They’re also bulky; it seems some­what ab­surd how much more bulky they are than a note­book of equiv­a­lently-sized pa­per. This is a se­ri­ous con­cern if you want to carry them around. (As a gen­eral rule, I’ve found that a binder feels roughly equiv­a­lent to one-size-larger note­book—a three-ring binder for 3x5 cards feels like car­ry­ing around a deck of 4x6 cards; a binder of A6 pa­per feels like a note­book of A5 pa­per; and so on.)

In­dex cards are of­ten kept in spe­cial boxes, which you can get. How­ever, I don’t like this so much? I want a more binder-like thing which I can eas­ily hold in my hands and flip through. Also, boxes are of­ten made to view cards in land­scape ori­en­ta­tion, but I pre­fer por­trait ori­en­ta­tion—so it’s hard to flip through things and read while they’re still in the box.

Cur­rently, I use the Sta­ples in­dex-cards-on-a-ring which put all the cards on a sin­gle ring, and pro­tect them with plas­tic cov­ers. How­ever, I re­place the metal rings (which I find harder to work with) with plas­tic rings. I also bought a va­ri­ety of note cards to try—you can try thicker/​thin­ner pa­per, col­ors, line grid, dot grid, etc. If you do this, you’ll need a hole punch, too. I recom­mend get­ting a “low force” hole punch; if you just go and buy the cheap­est hole punch you can find, it’ll prob­a­bly be pretty ter­rible. You want to be fairly con­sis­tent with where you punch the holes, but, that wasn’t as im­por­tant as I ex­pected (it doesn’t mat­ter as much with a one-ring binder in con­trast to a three-ring, since you’re not try­ing to get holes to line up with each other).

I en­joy the ring stor­age method, be­cause it makes cards re­ally easy to flip through, and I can work on sev­eral cards at once by splay­ing them out (which means I don’t lose my place when I de­cide to make a new card or make a note on a differ­ent one, and don’t have to take things out of sort or­der to work with them).

Deck Architecture

I don’t keep the cards perfectly sorted all the time. In­stead, I di­vide things up into sorted and not-yet-sorted:

(Blue in this image mean “writ­ten on”—they’re all ac­tu­ally white ex­cept for the yel­low di­vider, al­though of course you could use col­ored cards if you like.)

Fetch Modi

As I write on blank cards, I just leave them where they are, rather than im­me­di­ately putting them into the sort or­der­ing. I sort them in later.

There is an ad­van­tage to this ap­proach be­yond the effi­ciency of sort­ing things all at once. The un­sorted cards are a phys­i­cal record of what I’m ac­tively work­ing on. Since cards are so small, work­ing on an idea al­most always means cre­at­ing new cards. So, I can eas­ily jump back into what­ever I was think­ing about last time I han­dled the binder of cards.

Un­less you have a spe­cific new idea you want to think about (in which case you start a new card, or, go find the most closely re­lated cards in your ex­ist­ing pile), there are ba­si­cally two ways to en­ter into your card deck: from the front, and from the back. The front is “top-down” (both liter­ally and figu­ra­tively), go­ing from big­ger ideas to smaller de­tails. It’s more breadth-first. You’re likely to no­tice an idea which you’ve been ne­glect­ing, and start a new branch from it. Start­ing from the back, on the other hand, is depth-first. You’re con­tin­u­ing to go deeper into a branch which you’ve already de­vel­oped some depth in.

Don’t sort too of­ten. The un­sorted cards are a valuable record of what you’ve been think­ing about. I’ve re­gret­ted sort­ing too fre­quently—it feels like I have to start over, find the in­ter­est­ing open ques­tions buried in my stack of cards all over again.

In the­ory, one could also move cards from sorted to un­sorted speci­fi­cally to re­mind one­self to work on those cards, but I haven’t re­ally used this tac­tic.

Split­ting & Deck Management

When I have much more than 100 filled cards on a ring, I sort all of the cards, and split the deck into two. (Look for a sen­si­ble place to split the tree into two—you want to avoid a deep branch be­ing split up into two sep­a­rate decks, as much as you can.) Load up the two new decks with 50ish blank cards each, and stick them on new rings.

Every­thing is still on one big ad­dress­ing sys­tem, so, it is a good idea to la­bel the two new binders with the ad­dress range within. I use blank stick­ers, which I put on the front of each ring binder. The la­bels serve both to keep lookup easy (I don’t want to be guess­ing about which binder cer­tain ad­dresses are in), and also, to re­mind me to limit the ad­dresses within a given deck.

For ex­am­ple, sup­pose this is my first deck of cards (so be­fore the split, it holds ev­ery­thing). Let’s say there are 30 cards un­der­neath “1”, 20 cards un­der­neath “2”, and then about 50 more cards to­tal, un­der the num­bers 3 through 14.

I would split this deck into a “1 through 2” deck, and a “3 through *” deck—the * mean­ing “any­thing”. You might think it would be “3 through 14”, but, when I make card 15, it would go in that deck. So at any time, you have one deck of cards with no up­per bound. On the other hand, when you are work­ing with the “1 − 2” deck, you don’t want to mis­tak­enly make a card 3; you’ve already got a card 3 some­where. You don’t want du­pli­cate ad­dresses any­where!

Cur­rently, I have 6 decks: 0 − 1.4, 1.5 − 1.*, 2 − 2.4, 2.5 − 2.*, 3, and 4 − 4.*. (I was fool­ish when I started my Zet­telkas­ten, and used the dec­i­mal sys­tem rather than the alphanu­meric sys­tem. I switched quickly, but all my top-level ad­dresses are still dec­i­mal. So, I have a lot of mixed-ad­dress cards, such as 1.3a1, 1.5.2a2, 2.6b4a, etc. As for why my num­bers start at 0 rather than 1, I’ll dis­cuss that in the “In­dex & Bibliog­ra­phy” sec­tion.)

I like to have the un­sorted/​blank “short-term mem­ory” sec­tion on ev­ery sin­gle deck, so that I can con­ve­niently start think­ing about stuff within that deck with­out grab­bing any­thing else. How­ever, it might also make sense to have only one “short-term mem­ory” in or­der to keep your­self more fo­cused (and so that there’s only one place to check when you want to re­mem­ber what you were re­cently work­ing on!).

Get­ting Started: Your First Card

Your first note doesn’t need to be any­thing im­por­tant—it isn’t as if ev­ery idea you put into your Zet­telkas­ten has to be “un­der­neath” it. Re­mem­ber, you aren’t try­ing to in­vent a good cat­e­gory sys­tem. Not ev­ery card has to look like a core idea with bul­let points which elab­o­rate on that idea, like my ex­am­ple in the pre­vi­ous sec­tion. You can just start writ­ing what­ever. In fact, it might be good if you make your first cards messy and unim­por­tant, just to make sure you don’t feel like ev­ery­thing has to be nicely or­ga­nized and highly sig­nifi­cant.

On the other hand, it might be im­por­tant to have a good start­ing point, if you re­ally want to give Zet­telkas­ten a chance.

I men­tioned that I knew I liked Zet­telkas­ten within the first 30 min­utes. I think it might be im­por­tant that when I sat down to try it, I had an idea I was ex­cited to work on. It wasn’t a nice solid math­e­mat­i­cal idea—it was a fuzzy idea, one which had been burn­ing in the back of my brain for a week or so, wait­ing to be born. It filled the frac­tal branches of a zettelkas­ten nicely, ex­pand­ing in ev­ery di­rec­tion.

So, maybe start with one of those ideas. Some­thing you’ve been strug­gling to ar­tic­u­late. Some­thing which hasn’t found a place in your lin­ear note­book.

Alright. That’s all I have to say about the ba­sics of Zet­telkas­ten. You can go try it now if you want, or keep read­ing. The rest of this doc­u­ment is about fur­ther ideas in note-tak­ing which have shaped the way I use Zet­telkas­ten. Th­ese may or may not be crit­i­cal fac­tors; I don’t know for sure why Zet­telkas­ten is such a pro­duc­tive sys­tem for me per­son­ally.

Note-Tak­ing Sys­tems I Have Known and Loved

I’m or­ga­niz­ing this sec­tion by my pre­vi­ous note-tak­ing sys­tems, but se­cretly, the main point is to con­vey a num­ber of note-tak­ing ideas which may have con­tributed to Zet­telkas­ten work­ing well for me. Th­ese ideas have seemed gen­er­ally use­ful to me—maybe they’ll be use­ful to you, even if you don’t end up us­ing Zet­telkas­ten in par­tic­u­lar.


Devel­op­ing Ideas

Firstly, and most im­por­tantly, I have been keep­ing idea books since mid­dle school. I think there’s some­thing very im­por­tant in the sim­ple idea of writ­ing reg­u­larly—I don’t have the refer­ence, but, I re­mem­ber read­ing some­one who de­scribed the day they first started keep­ing a di­ary as the day they first woke up, started re­flec­tively think­ing about their re­la­tion­ship with the world. Here’s a some­what similar quote from a Zet­telkas­ten blog:

Dur­ing the time span­ning Nov. 2007–Jan. 2010, I filled 11 note books with ideas, to-do lists, ram­blings, di­ary en­tries, draw­ings, and wor­ries.
Look­ing back, this is about the time I started to live con­sciously. I guess keep­ing a jour­nal helped me “wake up” from some kind of teenage slum­ber.

I never got into au­to­bi­o­graph­i­cal di­ary-style writ­ing, per­son­ally, in­stead writ­ing about ideas I was hav­ing. Still, things were in a very “nar­ra­tive” for­mat—the ideas were a drama, a back-and-forth, a dance of re­join­ders. There was some math—pages filled with equa­tions—but only af­ter a great deal of (very) in­for­mal de­vel­op­ment of an idea.

As a re­sult, “elab­o­rate on an idea” /​ “keep go­ing” seems like a prim­i­tive op­er­a­tion to me—and, speci­fi­cally, a prim­i­tive op­er­a­tion which in­volves pa­per. (I can’t trans­late the same think­ing style to con­ver­sa­tion, not com­pletely.) I’m sure that there is a lot to un­pack, but for me, it just feels nat­u­ral to keep de­vel­op­ing ideas fur­ther.

So, when I say that the Zet­telkas­ten card 1b2 “elab­o­rates on” the card 1b, I’m call­ing on the long ex­pe­rience I’ve had with idea books. I don’t know if it’ll mean the same thing for you.

Here’s my in­com­plete at­tempt to con­vey some of what it means.

When I’m writ­ing in an idea book, I spend a lot of time try­ing to clearly ex­plain ideas un­der the (of­ten false) as­sump­tion that I know what I’m talk­ing about. There’s an imag­i­nary au­di­ence who knows a lot of what I’m talk­ing about, but I have to ex­plain cer­tain things. I can’t get away with leav­ing im­por­tant terms un­defined—I have to es­tab­lish any­thing I feel less than fully con­fi­dent about. For ex­am­ple, the defi­ni­tion of a Bayesian net­work is some­thing I can as­sume my “au­di­ence” can look up on wikipe­dia. How­ever, if I’m less than to­tally con­fi­dent in the con­cept of d-sep­a­ra­tion, I have to ex­plain it; es­pe­cially if it is im­por­tant to the ar­gu­ment I hope to make.

Once I’ve es­tab­lished the terms, I try to ex­plain the idea I was hav­ing. I spend a lot of time star­ing off into space, not re­ally know­ing what’s go­ing on in my head ex­actly, but with a sense that there’s a sim­ple point I’m try­ing to make, if only I could see it. I si­mul­ta­neously feel like I know what I want to say (if only I could find the words), and like I don’t know what it is—af­ter all, I haven’t ar­tic­u­lated it yet. Gen­er­ally, I can pick up where I left off with a par­tic­u­lar thought, even af­ter sev­eral weeks—I can glance at what I’ve writ­ten so far, and get right back to star­ing at the wall again, try­ing to ar­tic­u­late the same un-ar­tic­u­lated idea.

If I start again in a differ­ent note­book (for ex­am­ple, switch­ing to writ­ing my thoughts on a com­puter), I have to ex­plain ev­ery­thing again. This au­di­ence doesn’t know yet! I can’t just pick up on a com­puter where I left off on pa­per. It’s like try­ing to pick up a con­ver­sa­tion in the mid­dle, but with a differ­ent per­son. This is sort of an­noy­ing, but of­ten good (be­cause re-ex­plain­ing things may hold sur­prises, as I no­tice new de­tails.)

Similarly, if I do a lot of think­ing with­out a note­book (maybe in a con­ver­sa­tion), I gen­er­ally have to “con­struct” my new po­si­tion from my old one. This has an un­for­tu­nate “freez­ing” effect on thoughts: there’s a lot of grav­ity to­ward the chain of thought wher­ever it is on the page. I tend to work on what­ever line of thought is most re­cent in my note­book, re­gard­less of any more im­por­tant or bet­ter ideas which have come along—es­pe­cially if the line of thought in the note­book isn’t yet at a con­clu­sive place. Some­times I put a scrib­ble in the note­book af­ter a line of thought, to in­di­cate ex­plic­itly that it no longer re­flects the state of my think­ing, to give my­self “per­mis­sion” to do some­thing else.

Once I’ve ar­tic­u­lated some point, then crit­i­cisms of the point of­ten be­come clear, and I’ll start writ­ing about them. I of­ten have a sense that I know how it’s go­ing to go a few steps ahead in this back-and-forth; a few cri­tiques and replies/​re­vi­sions. Espe­cially if the ideas are flow­ing faster than I can write them down. How­ever, it is im­por­tant to ac­tu­ally write things down, be­cause they of­ten don’t go quite as I ex­pect.

If an idea seems to have reached a nat­u­ral con­clu­sion, in­clud­ing all the cri­tiques/​replies which felt im­por­tant enough to write, I’ll of­ten write a list of “fu­ture work”: any open ques­tions I can think of, ap­pli­ca­tions, de­tails which are im­por­tant but not so im­por­tant that I want to write about them yet, etc. At this point, it is usu­ally time to write the idea up for a real au­di­ence, which will re­quire more de­tail and re­fine the idea yet fur­ther (pos­si­bly de­stroy­ing it, or chang­ing it sig­nifi­cantly, as I of­ten find a crit­i­cal flaw when I try to write an idea up for con­sump­tion by oth­ers).

If I don’t have any par­tic­u­lar idea I’m de­vel­op­ing, I may start fresh with a men­tal mo­tion like “OK, ob­vi­ously I know how to solve ev­ery­thing” and write down the grand solu­tion to ev­ery­thing, start­ing big-pic­ture and con­tin­u­ing un­til I get stuck. Or, in­stead, I might make a bul­leted list free-as­so­ci­at­ing about what I think the in­ter­est­ing prob­lems are—the things I don’t know how to do.


The next ad­vance in my idea notes was work­flowy. I still love the sim­plic­ity of work­flowy, even though I have moved on from it.

For those un­familar, Work­flowy is an out­lin­ing tool. I was un­fa­mil­iar with the idea be­fore Work­flowy in­tro­duced it to me. Word pro­ces­sors gen­er­ally sup­port nested bul­leted lists, but the page-like for­mat of a word pro­ces­sor limits the depth such lists can go, and it didn’t re­ally oc­cur to me to use these as a pri­mary mode of writ­ing. Work­flowy doesn’t let you do any­thing but this, and it pro­vides enough fea­tures to make it ex­tremely con­ve­nient and nat­u­ral.

Non­lin­ear Ideas: Branch­ing Development

Work­flowy in­tro­duced me to the pos­si­bil­ity of non­lin­ear for­mats for idea de­vel­op­ment. I’ve already dis­cussed this to some ex­tent, since it is also one of the main ad­van­tages of Zet­telkas­ten over or­di­nary note­books.

Sud­denly, I could con­tinue a thread any­where, rather than always pick­ing it up at the end. I could sketch out where I ex­pected things to go, with an out­line, rather than keep­ing all the points I wanted to hit in my head as I wrote. If I got stuck on some­thing, I could write about how I was stuck nested un­der­neath what­ever para­graph I was cur­rently writ­ing, but then col­lapse the meta-thoughts to be in­visi­ble later—so the over­all nar­ra­tive doesn’t feel in­ter­rupted.

In con­trast, writ­ing in pa­per note­books forces you to choose con­sciously that you’re done for now with a topic if you want to start a new one. Every new para­graph is like choos­ing a sin­gle fork in a twist­ing maze. Work­flowy al­lowed me to take them all.

What are Chil­dren?

I’ve seen peo­ple hit a block right away when they try to use work­flowy, be­cause they don’t know what a “child node” is.

  • Here’s a node. It could be a para­graph, ex­press­ing some thought. It could also be a ti­tle.

    • Here’s a child node. It could be a com­ment on the thought—an aside, a cri­tique, what­ever. It could be some­thing which goes un­der the head­ing.

  • Here’s a sibling node. It could be the next para­grapt in the “main thrust” of an ar­gu­ment. It could be an un­re­lated point un­der the same su­per-point ev­ery­thing is un­der.

As with Zet­telkas­ten, my ad­vice is to not get too hung up on this. A child is sort of like a com­ment; a par­en­thet­i­cal state­ment or a foot­note. You can con­tinue the main thrust of an ar­gu­ment in sibling nodes—just like writ­ing an or­di­nary se­quence of para­graphs in a word pro­ces­sor.

You can also or­ga­nize things un­der head­ings. This is es­pe­cially true if you wrote a sketchy out­line first and then filled it in, or, if you have a lot of ma­te­rial in Work­flowy and had to or­ga­nize it. The “up­per on­tol­ogy” of my work­flowy is mostly ti­tle-like, sin­gle words or short noun phrases. As you get down in, bul­lets start to be sen­tences and para­graphs more of­ten.

Ob­vi­ously, all of this can be ap­plied to Zet­telkas­ten to some ex­tent. The biggest differ­ence is that “up­per-level” cards are less likely to just be cat­e­gory ti­tles; and, you can’t re­ally or­ga­nize things into nice cat­e­gories af­ter-the-fact be­cause the ad­dresses in Zet­telkas­ten are fixed—you can’t change them with­out break­ing links. You can use redi­rect cards if you want to re­or­ga­nize things, ac­tu­ally, but I haven’t done that very much in prac­tice. Some­thing which has worked for me to some ex­tent is to re­or­ga­nize things in the in­dexes. Once an in­dex is too much of a big flat list, you can cluster en­tries into sub­jects. This new list­ing can be added as a child to the pre­vi­ous in­dex, keep­ing the his­tor­i­cal record; or, pos­si­bly, re­place the old in­dex out­right. I dis­cuss this more in the sec­tion on in­dex­ing.

Build­ing Up Ideas over Long Time Periods

My idea books let me build up ideas over time to a greater ex­tent than my peers who didn’t keep similar jour­nals. How­ever, be­cause the lin­ear for­mat forces you to switch top­ics in a se­rial man­ner and “start over” when you want to re­sume a sub­ject, you’re mostly re­stricted to what you can keep in your head. Your note­books are a form of in­for­ma­tion stor­age, and you can go back and re-read things, but only if you re­mem­ber the rele­vant item to go back and re-read.

Work­flowy al­lowed me to build up ideas to a greater de­gree, in­cre­men­tally adding thoughts un­til cas­cades of un­der­stand­ing changed my over­all view.

Plac­ing a New Idea

Be­cause you’ve got all your ideas in one big out­line, you can add in lit­tle ideas eas­ily. Work­flowy was easy enough to ac­cess via my smart­phone (though they didn’t have a proper app at the time), so I could jot down an idea as I was walk­ing to class, wait­ing for the bus, etc. I could eas­ily nav­i­gate to the right lo­ca­tion, at least, if I had or­ga­nized the over­all struc­ture of the out­line well. Writ­ing one lit­tle idea would usu­ally get more flow­ing, and I would add sev­eral points in the same lo­ca­tion on the tree, or in nearby lo­ca­tions.

This idea of jot­ting down ideas while you’re out and about is very im­por­tant. If you feel you don’t have enough ideas (be it for re­search, for writ­ing fic­tion, for art—what­ever) my first ques­tion would be whether you have a good way to jot down lit­tle ideas as they oc­cur to you.

The fact that you’re forced to some­how fit all ideas into one big tree is also im­por­tant. It makes you or­ga­nize things in ways that are likely to be use­ful to you later.

Or­ga­niz­ing Over Time

The sec­ond re­ally nice thing work­flowy did was al­low me to go back and re­or­ga­nize all the lit­tle ideas I had jot­ted down. When I sat down at a com­puter, I could take a look at my tree over­all and see how well the cat­e­go­riza­tion fit. This mostly took the form of small im­prove­ments to the tree struc­ture over time. Even­tu­ally, a cas­cade of small fixes turned into a ma­jor re­or­ga­ni­za­tion. At that point, I felt I had re­ally learned some­thing—all the in­cre­men­tal progress built up into an over­all shift in my un­der­stand­ing.

Again, this isn’t re­ally pos­si­ble in pa­per-based Zet­telkas­ten—the ad­dress sys­tem is fixed. How­ever, as I men­tioned be­fore, I’ve had some suc­cess do­ing this kind of re­or­ga­ni­za­tion within the in­dexes. It doesn’t mat­ter that the ad­dresses of the cards are fixed if the way you ac­tu­ally find those ad­dresses is muta­ble.

Limi­ta­tions of Workflowy

Even­tu­ally, I no­ticed that I had a big pile of ideas which I hadn’t re­ally de­vel­oped. I was jot­ting down ideas, sure. I was fit­ting them into an in­creas­ingly co­he­sive over­all pic­ture, sure. But I wasn’t do­ing any­thing with them. I wasn’t writ­ing pages and pages of de­tails and cri­tique.

It was around this time that I re­al­ized I had gone more than three years with­out us­ing a pa­per note­book very sig­nifi­cantly. I started writ­ing on pa­per again. I re­al­ized that there were all these habits of think­ing which were tied to pa­per for me, and which I didn’t re­ally ac­cess if I didn’t have a nice note­book and a nice pen—the force of the long-prac­ticed as­so­ci­a­tions. It was like wak­ing up in­tel­lec­tu­ally af­ter hav­ing gone to sleep for a long time. I started to re­mem­ber high­school. It was a weird time. Any­way...


The next thing I tried was Dy­nal­ist.

The main ad­van­tage of Dy­nal­ist over Work­flowy is that it takes a fea­ture-rich rather than min­i­mal­is­tic ap­proach. I like the clean aes­thet­ics of Work­flowy, but… even­tu­ally, there’ll be some crit­i­cal fea­ture Work­flowy just doesn’t provide, and you’ll want to make the jump to Dy­nasilt. I use hardly any of the ex­tra fea­tures of Dy­nal­ist, but the ones I do use, I need. For me, it’s mostly the LaTeX sup­port.

Another thing about Dy­nal­ist which felt very differ­ent for me was the file sys­tem. Work­flowy forces you to keep ev­ery­thing in one big out­line. Dy­nal­ist lets you cre­ate many out­lines, which it treats as differ­ent files; and, you can or­ga­nize them into fold­ers (re­cur­sively). Tech­ni­cally, that’s just an­other tree struc­ture. In terms of UI, though, it made nav­i­ga­tion much eas­ier (be­cause you can eas­ily ac­cess a de­sired file through the file pane). Psy­cholog­i­cally, it made me much more will­ing to start fresh out­lines rather than add to one big one. This was both good and bad. It meant my ideas were less an­chored in one big tree, but it even­tu­ally re­sulted in a big, di­s­or­ga­nized pile of notes.

I did learn my les­son from Work­flowy, though, and set things up in my Dy­nal­ist such that I ac­tu­ally de­vel­oped ideas, rather than just col­lect­ing scraps for­ever.

Tem­po­rary Notes vs Or­ga­nized Notes

I or­ga­nized my Dy­nal­ist files as fol­lows:

  • A “log” file, in which I could write what­ever I was think­ing about. This was or­ga­nized by date, al­though I would of­ten go back and elab­o­rate on things from pre­vi­ous dates.

  • A “todo” file, where I put links to items in­side “log” which I speci­fi­cally wanted to go back and think more about. I would pe­ri­od­i­cally sort the todo items to re­flect my pri­ori­ties. This gave me a list of im­por­tant top­ics to draw from when­ever I wasn’t sure what I wanted to think about.

  • A bunch of other di­s­or­ga­nized files.

This sys­tem wasn’t great, but it was a whole lot bet­ter at ac­tu­ally de­vel­op­ing ideas than the way I kept things or­ga­nized in Work­flowy. I had re­al­ized that lock­ing ev­ery­thing into a unified tree struc­ture, while good for the pur­pose of slowly im­prov­ing a large on­tol­ogy which or­ga­nized a lot of lit­tle thoughts, was keep­ing me from just writ­ing what­ever I was think­ing about.

Dan Sheffler (whose es­says I’ve already cited sev­eral times in this writeup) writes about re­al­iz­ing that his note-tak­ing sys­tem was si­mul­ta­neously try­ing to im­ple­ment two differ­ent goals: an or­ga­nized long-term mem­ory store, and “en­gage­ment notes” which are writ­ten to clar­ify think­ing and have a more stream-of-con­scious­ness style. My “log” file was es­sen­tially en­gage­ment notes, and my “todo” file was the long-term mem­ory store.

For some peo­ple, I think an es­sen­tial part of Zet­telkas­ten is the dis­tinc­tion be­tween tem­po­rary and per­ma­nent notes. Tem­po­rary notes are the di­s­or­ga­nized stream-of-con­scious­ness notes which Sheffler calls en­gage­ment notes. Tem­po­rary notes can also in­clude all sorts of other things, such as todo lists which you make at the start of the day (and which only ap­ply to that day), shop­ping lists, etc. Tem­po­rary notes can be kept in a lin­ear for­mat, like a note­book. Pe­ri­od­i­cally, you re­view the tem­po­rary notes, putting the im­por­tant things into Zet­telkas­ten.

In Tak­ing Smart Notes, Luh­mann is de­scribed as trans­fer­ring the im­por­tant thoughts from the day into Zet­tel ev­ery evening. Sheffler, on the other hand, keeps a gap of at least 24 hours be­tween tak­ing down en­gage­ment notes and de­cid­ing what be­longs in the long-term store. A gap of time al­lows the ini­tial ex­cite­ment over an idea to pass, so that only the things which still seem im­por­tant the next day get into long-term notes. He also points out that this sys­tem en­forces a small amount of spaced rep­e­ti­tion, mak­ing it more likely that con­tent is re­called later.

As for my­self, I mostly write di­rectly into my Zet­telkas­ten, and I think it’s pretty great. How­ever, I do find this to be difficult/​im­pos­si­ble when tak­ing quick notes dur­ing a con­ver­sa­tion or a talk – when I try, then the re­sult­ing con­tent in my Zet­telkas­ten seems pretty use­less (ie, I don’t come back to it and fur­ther de­velop those thoughts). So, I’ve started to carry a note­book again for those tem­po­rary notes.

I cur­rently think of things like this:


Th­ese are the sort of small poin­t­ers to ideas which you can write down while walk­ing, wait­ing for the bus, etc. The idea is stated very sim­ply—per­haps in a sin­gle word or a short phrase. A sen­tence at most. You might for­get what it means af­ter a week, es­pe­cially if you don’t record the con­text well. The first thing to re­al­ize about jots is to cap­ture them at all, as already dis­cussed. The sec­ond thing is to cap­ture them in a place where you will be able to de­velop them later. I used to carry around a small pocket note­book for jots, af­ter I stopped us­ing Work­flowy reg­u­larly. My plan was to re­view the jots when­ever I filled a note­book, putting them in more long-term stor­age. This never hap­pened: when I filled up a note­book, un­pack­ing all the jots into some­thing mean­ingful just seemed like too huge a task. It works bet­ter for me to jot things into per­ma­nent stor­age di­rectly, as I did with Work­flowy. I pro­cras­ti­nate too much on turn­ing tem­po­rary notes into long term notes, and the tem­po­rary notes be­come mean­ingless.


A gloss is a para­graph ex­plain­ing the point of a jot. If a jot is the ti­tle of a Zet­telkas­ten card, a gloss is the first para­graph (of­ten writ­ten in a dis­tinct color). This gives enough of an idea that the thought will not be lost if it is left for a few weeks (per­haps even years, de­pend­ing). Writ­ing a gloss is usu­ally easy, and do­ing so is of­ten enough to get the ideas flow­ing.


This is the kind of writ­ing I de­scribed in the ‘note­books’ sec­tion. An idea is fleshed out. This kind of writ­ing is of­ten still com­pre­hen­si­ble years later, al­though it isn’t guaran­teed to be.


This is the kind of writ­ing which is pub­lish­able. It nails the idea down. There’s not re­ally any end to this—you can imag­ine ex­pand­ing some­thing from a blog post, to an aca­demic pa­per, to a book, and fur­ther, with in­creas­ing lev­els of de­tail, gen­tle ex­po­si­tion, for­mal rigor—but to a first ap­prox­i­ma­tion, any­way, you’ve elimi­nated all the con­tra­dic­tions, stated the mo­ti­vat­ing con­text ac­cu­rately, etc.

I called the last item “re­fine­ment” rather than “com­mu­ni­ca­tion” be­cause, re­ally, you can com­mu­ni­cate your ideas at any of these stages. If some­one shares a lot of con­text with you, they can un­der­stand your jots. That’s re­ally difficult, though. More likely, a re­search part­ner will un­der­stand your glosses. Devel­op­ment will be un­der­stand­able to some­one a lit­tle more dis­tant, and so on.

At Long Last, Zettelkasten

I’ve been ham­mer­ing home the idea of “lin­ear” vs “non­lin­ear” for­mats as one of the big ad­van­tages of Zet­telkas­ten. But work­flowy and dy­nal­ist both al­low non­lin­ear writ­ing. Why should you be in­ter­ested in Zet­telkas­ten? Is it any­thing more than a way to im­ple­ment work­flowy-like writ­ing for a pa­per for­mat?

I’ve said that (at least for me) there’s some­thing ex­tra-good about Zet­telkas­ten which I don’t re­ally un­der­stand. But, there are a cou­ple of im­por­tant el­e­ments which make Zet­telkas­ten more than just pa­per work­flowy.

  • Hier­ar­chy Plus Cross-Links: A re­peated theme across knowl­edge for­mats, in­clud­ing wikipe­dia and text­books, is that you want both a hi­er­ar­chi­cal or­ga­ni­za­tion which makes it easy to get an overview and find things, and also a “cross-refer­ence” type ca­pa­bil­ity which al­lows re­lated con­tent to be linked—cre­at­ing a het­er­ar­chi­cal web. I men­tioned at the be­gin­ning that Zet­telkas­ten forced me to cre­ate cross-links much more than I oth­er­wise would, due to the use of small note-cards. Work­flowy has “hi­er­ar­chy” down, but it has some­what poor “cross-link” ca­pa­bil­ity. It has tags, but a tag sys­tem is not as pow­er­ful as hy­per­text. Be­cause you can link to in­di­vi­d­ual nodes, it’s pos­si­ble to use hy­per­text cross-links—but the pro­cess is awk­ward, since you have to get the link to the node you want. Dy­nal­ist is sig­nifi­cantly bet­ter in this re­spect—it has an easy way to cre­ate a link to any­thing by search­ing for it (with­out leav­ing the spot you’re at). But it lacks the wiki-style “magic link” ca­pa­bil­ity, cre­at­ing a new page when you make a link which has no tar­get. Roam, how­ever, pro­vides this fea­ture.

  • Atomic­ity: The idea of cre­at­ing pages or­ga­nized around a sin­gle idea (again, an idea re­lated to wikis). This is pos­si­ble in Dy­nal­ist, but Zet­telkas­ten prac­ti­cally forces it upon you, which for me was re­ally good. Again, Roam man­ages to en­courage this style.

Zet­telkas­ten, Part 2: Fur­ther Advice

Card Layout

My cards of­ten look some­thing like this:

I’m left handed, so you may want to flip all of this around if you’re right handed. I use the ring binder “back­wards” from the in­tended con­figu­ra­tion (the punched hole would usu­ally be on the left, rather than the right). Also, I pre­fer por­trait rather than land­scape. Most peo­ple pre­fer to use 3x5 cards in land­scape, I sup­pose.

Any­way, not ev­ery card will look ex­actly like the above. A card might just con­tain a bunch of free-writ­ing, with no bul­leted list. Or it might only con­tain a bul­leted list, with no blurb at the be­gin­ning. What­ever works. I think my lay­out is close to Luh­mann’s and close to com­mon ad­vice—but if you try to copy it re­li­giously, you’ll prob­a­bly feel like Zet­telkas­ten is awk­ward and re­stric­tive.

The only ab­solutely nec­es­sary thing is the ad­dress. The ad­dress is the first thing you write on a new card. You don’t ever want a card to go with­out an ad­dress. And it should be in a stan­dard lo­ca­tion, so that it is re­ally easy to look through a bunch of cards for one with a spe­cific ad­dress.

Don’t feel bad if you start a card and leave it mostly blank for­ever. Maybe you thought you were go­ing to elab­o­rate an idea, so you made a new card, but it’s got noth­ing but an ad­dress. That’s ok. Maybe you will fill it later. Maybe you won’t. Don’t worry about it.

Mostly, a thought is con­tinued through elab­o­ra­tion on bul­let points. I might write some­thing like “cont. 1.1a1a” at the bot­tom of the card if there’s an­other card that’s re­ally a di­rect con­tinu­a­tion, though. (Ac­tu­ally, I don’t write “cont.”; I just write the down ar­row, which means the same thing.) If so, I’d write “see 1.1a1” in the up­per left hand cor­ner, to in­di­cate that 1.1a1a prob­a­bly doesn’t make much sense on its own with­out con­sult­ing 1.1a1 -- moreso than usual for child cards. (Ac­tu­ally, I’d write an­other down ar­row rather than “see”, mir­ror­ing the down ar­row on the pre­vi­ous card—this in­di­cates the di­rect-con­tinu­a­tion re­la­tion­ship.)

In the illus­tra­tion, I wrote links [in square brack­ets]. The truth is, I of­ten put them in full rec­t­an­gu­lar boxes (to make them stand out more), al­though not always. Some­times I put them in paren­the­ses when I’m us­ing them more as a noun, as in: “I think pizza (12a) might be rele­vant to pasta. [14x5b]” In that ex­am­ple, “(12a)” is the card for pizza. “[14x5b]” is a card con­tin­u­ing the whole thought “pizza might be rele­vant to pasta”. So paren­the­ses-vs-box is sort of like top-cor­ner-vs-bot­tom, but for an in­di­vi­d­ual line rather than a whole card.

Use of Color

The col­ors are true to my writ­ing as well. For a long time, I wanted to try writ­ing with multi-color click pens, be­cause I knew some peo­ple found them very use­ful; but, I was un­able to find any which satis­fied my (ex­cep­tion­ally picky) taste. I don’t gen­er­ally go for ball-point pens; they aren’t smooth enough. I pre­fer to write with felt-tip draw­ing pens or similar. I also pre­fer very fine tips (as a con­se­quence of prefer­ring my writ­ing to be very small, as I men­tioned pre­vi­ously) -- al­though I’ve also found that the ap­pro­pri­ate line width varies with my men­tal state and with the sub­ject mat­ter. Fine lines are bet­ter for fine de­tails, and for en­er­getic men­tal states; broad lines are bet­ter for loose free-as­so­ci­a­tion and brain­storm­ing, and for tired men­tal states.

In any case, a friend recom­mended the Hi-Tec C Co­leto, a multi-color click pen which feels as smooth as felt-tip pens usu­ally do (al­most). You can buy what­ever col­ors you want, and they’re available in a va­ri­ety of line-widths, so you can cus­tomize it quite a bit.

At first I just used differ­ent col­ors hap­haz­ardly. I figured I would even­tu­ally set­tle on mean­ings for col­ors, if I just used what­ever felt ap­pro­pri­ate and ex­per­i­mented. Mostly, that meant that I switched col­ors to in­di­cate a change of topic, or used a differ­ent color when I went back and an­no­tated some­thing (which re­ally helps read­abil­ity, by the way—black writ­ing with a bunch of black an­no­ta­tions scrib­bled next to it or be­tween lines is hard to read, com­pared to pur­ple writ­ing with or­ange an­no­ta­tions, or what­ever!). When I switched to Zet­telkas­ten, though, I got more sys­tem­atic with my use of color.

I roughly fol­low Lion Kim­bro’s ad­vice about col­ors, from How to Make a Com­plete Map of Every Thought you Think:

Now lets talk about color.
Your pen has four col­ors: Red, Green, Blue, and Black
You will want to con­nect mean­ing with each color.
Here’s my as­so­ci­a­tions:
RED: Er­ror, Warn­ing, Correction
BLUE: Struc­ture, Di­a­gram, Pic­ture, Links, Keys (in key-value pairs)
GREEN: Meta, Defi­ni­tion, Nam­ing, Brief An­no­ta­tion, Glyphs
BLACK: Main Content
I also use green to clar­ify sloppy writ­ing later on. Blue is for Keys, Black is for val­ues.
I hope that’s self-ex­plana­tory.
If you make a cor­rec­tion, put it in red. Page num­bers are blue. If you draw a di­a­gram, make it blue. Main con­tent in black.
Sup­pose you make a di­a­gram: Start with a big blue box. Put the di­a­gram in the box. (Or the other way around- make the di­a­gram, than the box around it.) Put some high­lighted con­tent in black. Want to define a word? Use a green cal­lout. Oops- there’s a prob­lem in the draw­ing- X it out in red, fol­lowed by the cor­rec­tion, in red.
Some times, I use black and blue to al­ter­nate em­pha­sis. Black and blue are the eas­iest to see.
If I’m an­no­tat­ing some text in the fu­ture, and the text is black, I’ll switch to us­ing blue for con­tent. Or vise versa.
Some an­no­ta­tions are red, if they are ma­jor cor­rec­tions.
Always re­mem­ber: Tol­er­ate er­rors. If your black has run out, and you don’t want to get up right away to fetch your backup pen, then just switch to blue. When the thoughts out, go get your backup pen.

The only big differ­ences are that I use brown in­stead of black in my pen, I tend to use red for ti­tles so that they stand out very clearly, and I use green for links rather than blue.

In­dex & Bibliog­ra­phy?


Tak­ing Smart Notes de­scribes two other kinds of cards: in­dexes, and biblio­graph­i­cal notes. I haven’t made those work for me very effec­tively, how­ever. Luh­mann, the in­ven­tor of Zet­telkas­ten, is de­scribed in­vent­ing Zet­telkas­ten as a way to or­ga­nize notes origi­nally made while read­ing. I don’t use it like that—I mainly use it for or­ga­niz­ing notes I make while think­ing. So bibliog­ra­phy isn’t of pri­mary im­por­tance for me.

(Ap­par­ently Um­berto Eco similarly ad­vises keep­ing idea notes and read­ing notes on sep­a­rate sets of in­dex cards.)


So I don’t miss the bibliog­ra­phy cards. (Maybe I will even­tu­ally.) On the other hand, I definitely need some sort of in­dex, but I’m not sure about the best way to keep it up to date. I only no­tice that I need it when I go look­ing for a par­tic­u­lar card and it is difficult to find! When that hap­pens, and I even­tu­ally find the card I wanted, I can jot down its ad­dress in an in­dex. But, it would be nice to some­how avoid this. So, I’ve ex­per­i­mented with some ideas. Here are some­one else’s thoughts on in­dex­ing (for a digi­tal zettelkas­ten).

List­ing As­sorted Cards

The first type of in­dex which I tried lists “im­por­tant” cards (cards which I re­fer to of­ten). I just have one of these right now. The idea is that you write a card’s name and ad­dress on this in­dex if you find that you’ve had difficulty lo­cat­ing a card and wished it had been listed in your in­dex. This sounds like it should be bet­ter than a sim­ple list of the top-level num­bered cards, since (as I men­tioned ear­lier) cards like 11a of­ten turn out to be more im­por­tant than cards like 11. Un­for­tu­nately, I’ve found this not to be the case. The prob­lem is that this kind of in­dex is too hard to main­tain. If I’ve just been strug­gling to find a card, my work­ing mem­ory is prob­a­bly already over-taxed with all the stuff I wanted to do af­ter find­ing that card. So I for­get to add it to the in­dex.

Topic Index

Some­times it also makes sense to just make a new top-level card on which you list ev­ery­thing which has to do with a par­tic­u­lar cat­e­gory. I have only done this once so far. It seems like this is the main mode of in­dex­ing which other peo­ple use? But I don’t like the idea that well.

List­ing Sibling Cards

When a card has enough chil­dren that they’re difficult to keep track of, I add a “zero” card be­fore all the other chil­dren, and this works as an in­dex. So, for ex­am­ple, card 2a might have chil­dren 2a1, 2a2, 2a3, … 2a15. That’s a lot to keep track of. So I add 2a0, which gets an en­try for 2a1-2a15, and any new cards added un­der 2a. It can also get an en­try for par­tic­u­larly im­por­tant de­scen­dants; maybe 2a3a1 is ex­tra im­por­tant and gets an en­try.

For cards like 2, whose chil­dren are alpha­bet­i­cal, you can’t re­ally use “zero” to go be­fore all the other chil­dren. I use “λ” as the “alpha­bet­i­cal zero”—I sort it as if it comes be­fore all the other let­ters in the alpha­bet. So, card “1λ” lists 1a, 1b, etc.

The most im­por­tant in­dex is the in­dex at 0, ei, the in­dex of all top-level num­bered cards. As I de­scribe in the “card lay­out” sec­tion, a card already mostly lists its own chil­dren—mean­ing that you don’t need to add a new card to serve this pur­pose un­til things get un­wieldy. How­ever, top-level cards have no par­ents to keep track of them! So, you prob­a­bly want an “ab­solute zero” card right away.

Th­ese “zero” cards also make it eas­ier to keep track of whether a card with a par­tic­u­lar ad­dress has been cre­ated yet. Every time you make a card, you add it to the ap­pro­pri­ate zero card; so, you can see right away what the next ad­dress available is. This isn’t the case oth­er­wise, es­pe­cially if your cards aren’t cur­rently sorted.

Kim­bro’s Mind Mapping

I’ve ex­per­i­mented with adapt­ing Lion Kim­bro’s sys­tem from How to Make a Com­plete Map of Every Thought You Think. After all, a com­plete map of ev­ery thought you think sounds like the perfect in­dex!

In my ter­minol­ogy, Lion Kim­bro keeps only jots—he was fo­cus­ing on col­lect­ing and map­ping, rather than de­vel­op­ing, ideas. Jots were col­lected into top­ics and sub-top­ics. When an area ac­cu­mu­lated enough jots, he would start a mind map for it. I won’t go into all his spe­cific map­ping tips (al­though they’re rele­vant), but ba­si­cally, imag­ine putting the ad­dresses of cards into clusters (on a new blank card) and then writ­ing “an­chor words” de­scribing the clusters.

You built your tree in an ini­tially “top-down” fash­ion, ex­pand­ing trees by adding in­creas­ingly-nested cards. You’re go­ing to build the map “bot­tom-up”: when a sub-tree you’re in­ter­ested in feels too large to quickly grasp, start a map. Let’s say you’re map­ping card 8b4. You might already have an in­dex of chil­dren at 8b4; if that’s the case, you can start with that. Also look through all the de­scen­dants of 8b4 and pick out whichever seem most im­por­tant. (If this is too hard, start by mak­ing maps for 8b4’s chil­dren, and re­turn to map­ping 8b4 later.) Draw a new mind map, and place it at 8b4a—it is part of the in­dex; you want to find it eas­ily when look­ing at the in­dex.

Now, the im­por­tant thing is that when you make a map for 8b, you can take a look at the map for 8b4, as well as any maps pos­sessed by other chil­dren of 8b. This means that you don’t have to go through all of the de­scen­dents of 8b (which is good, be­cause there could be a lot). You just look at the maps, which already give you an overview. The map for 8b is go­ing to take the most im­por­tant-seem­ing el­e­ments from all of those sub-maps.

This al­lows im­por­tant things to trickle up to the top. When you make a map at 0, you’ll be get­ting all the most im­por­tant stuff from deep sub-trees just by look­ing at the maps for each top-level num­bered card.

The cat­e­gories which emerge from map­ping like this can be com­pletely differ­ent from the con­cepts which ini­tially seeded your top-level cards. You can make new top-level cards which cor­re­spond to these cat­e­gories if you want. (I haven’t done this.)

Now, when you’re look­ing for some­thing, you start at your top-level map. You look at the clusters and likely have some ex­pec­ta­tion about where it is (if the ad­dress isn’t some­where on your top-level map already). You fol­low the ad­dresses to fur­ther maps, which give fur­ther clusters of ad­dresses, un­til you land in a tree which is small enough to nav­i­gate with­out maps.

I’ve de­scribed all of this as if it’s a one-time op­er­a­tion, but of course you keep adding to these maps, and re-draw up­dated maps when things don’t fit well any more. If a map lives at 8b40a, then the up­dated maps can be 8b40b, 8b40c, and so on.You can keep the old maps around as a his­tor­i­cal record of your shift­ing con­cep­tual clusters.

Keep­ing Mul­ti­ple Zettelkasten

A note sys­tem like Zet­telkas­ten (or work­flowy, dy­nal­ist, ev­er­note, etc) is sup­posed to stick with you for years, grow­ing with you and be­com­ing a repos­i­tory for your ideas. It’s a big com­mit­ment.

It’s difficult to op­ti­mize note-tak­ing if you think of it that way, though. You can’t ex­per­i­ment if you have to look be­fore you leap. I would have never tried Zet­telkas­ten if I thought I was com­mit­ting to try it as my “next sys­tem”—I didn’t think it would work.

Similarly, I can’t op­ti­mize my Zet­telkas­ten very well with that at­ti­tude. A Zet­telkas­ten is sup­posed to be one repos­i­tory for ev­ery­thing—you’re not sup­posed to start a new one for a new pro­ject, for ex­am­ple. But, I have sev­eral Zet­telkas­ten, to test out differ­ent for­mats: differ­ent sizes of card, differ­ent binders. It is still difficult to give al­ter­na­tives a fair shake, be­cause my two main Zet­telkas­ten have built up mo­men­tum due to the con­tent I keep in them.

I use a sys­tem of cap­i­tal let­ters to cross refer­ence be­tween my Zet­telkas­ten. For ex­am­ple, my main 3x5 Zet­telkas­ten is “S” (for “small”). I have an­other Zet­telkas­ten which is “M”, and also an “L”. When refer­enc­ing card 1.1a within S, I just call it 1.1a. If I want to re­fer to it from a card in M, I call it S1.1a in­stead. And so on.

Ap­par­ently Luh­mann did some­thing similar, start­ing a new Zet­telkas­ten which oc­ca­sion­ally referred to his first.

How­ever, keep­ing mul­ti­ple Zet­telkas­ten for spe­cial top­ics is not nec­es­sar­ily a good idea. Be­ware fixed cat­e­gories. The dan­ger is that cat­e­gories limit what you write, or, be­come less ap­pro­pri­ate over time. I’ve tried spe­cial-topic note­books in the past, and while it does some­times work, I of­ten end up con­flicted about where to put some­thing. (Granted, I have a similar con­flict about where to put things in my sev­eral omni-topic Zet­telkas­ten, but mostly the 3x5 sys­tem I’ve de­scribed here has won out—for now.)

On the other hand, I sus­pect it’s fine to cre­ate spe­cial topic zettelkas­ten for “very differ­ent” things. Creat­ing a new zettelkas­ten be­cause you’re writ­ing a new book is prob­a­bly bad—al­though it’ll work fine for the goal of or­ga­niz­ing ma­te­rial for writ­ing books, it means your next book idea isn’t com­ing from Zet­telkas­ten. (Zet­telkas­ten should con­tain/​ex­tend the thought pro­cess which gen­er­ates book ideas in the first place, and it can’t do that very well if you have to have a spe­cific book idea in or­der to start a zettelkas­ten about it.) On the other hand, I sus­pect it is OK to keep a sep­a­rate Zet­telkas­ten for fic­tional cre­ative writ­ing. Fac­tual ideas can spark ideas for fic­tion, but, the two are suffi­ciently differ­ent “modes” that it may make sense to keep them in phys­i­cally sep­a­rate col­lec­tions.

The idea of us­ing an ex­tended ad­dress sys­tem to make refer­ences be­tween mul­ti­ple Zet­telkas­ten can also be ap­plied to ad­dress other things, out­side of your Zet­telkas­ten. For ex­am­ple, you might want come up with a way of adding ad­dresses to your old note­books so that you can re­fer to them eas­ily. (For ex­am­ple, “note­book num­ber: page num­ber” could work.)

Should You Trans­fer Old Notes Into Zet­telkas­ten?

Re­lat­edly, since Zet­telkas­ten ideally be­comes a library of all the things you have been think­ing about, it might be tempt­ing to try and trans­fer ev­ery­thing from your ex­ist­ing notes into Zet­telkas­ten.

(A lot of read­ers may not even be tempted to do this, given the amount of work it would take. Yet, those more se­ri­ous about note sys­tems might think this is a good idea—or, might be too afraid to try Zet­telkas­ten be­cause they think they’d have to do this.)

I think trans­fer­ring older stuff into Zet­telkas­ten can be use­ful, but, try­ing to make it hap­pen right away as one big pro­ject is most likely not worth it.

  • It’s true that part of the use­ful­ness of Zet­telkas­ten is the in­ter­con­nected web of ideas which builds up over time, and the “high-sur­face-area” for­mat which makes it easy to branch off any part. How­ever, not all the pay­off is long-term: it should also be use­ful in the mo­ment. You’re not only writ­ing notes be­cause they may help you de­velop ideas in the fu­ture; the act of writ­ing the notes should be helping you de­velop ideas now.

  • You should prob­a­bly only spend time putting ideas into Zet­telkas­ten if you’re ex­cited about fur­ther de­vel­op­ing those ideas right now. You should not just be copy­ing over ideas into Zet­telkas­ten. You should be im­prov­ing ideas, think­ing about where to place them in your ad­dress hi­er­ar­chy, in­ter­link­ing them with other ideas in your Zet­telkas­ten via ad­dress links, and tak­ing notes on any new ideas sparked by this pro­cess. Try­ing to put all your old notes into Zet­telkas­ten at once will likely make you feel hur­ried and un­will­ing to de­velop things fur­ther as you go. This will re­sult in a pile of mediocre notes which will ul­ti­mately be less use­ful.

  • I men­tioned the breadth-first vs depth-first dis­tinc­tion ear­lier. Put­ting all of your old notes into Zet­telkas­ten is an ex­tremely breadth-first strat­egy, which likely doesn’t give you enough time to go deep into fur­ther de­vel­op­ing any one idea.

What about the dream of hav­ing all your notes in one beau­tiful for­mat? Well, it is true that old notes in differ­ent for­mats may be harder to find, since you have to re­mem­ber what for­mat the note you want was writ­ten in, or check all your old note sys­tems to find the note you want. I think it just isn’t worth the cost to fix this prob­lem, though, es­pe­cially since you should prob­a­bly try many differ­ent sys­tems to find a good one that works for you, and you can’t very well port all your notes to each new sys­tem.

Zet­telkas­ten should be an over­all im­prove­ment com­pared to a nor­mal note­book—if it isn’t, you have no busi­ness us­ing it. Ad­ding a huge up-front cost of trans­fer­ring notes un­der­mines that. Just pick Zet­telkas­ten up when you want to use it to de­velop ideas fur­ther.

Depth-first vs Breadth-first

Speak­ing of depth-first vs breadth-first, how should you bal­ance those two modes?

Luck­ily, this prob­lem has some rele­vant com­puter sci­ence the­ory be­hind it. I tend to think of it in terms of iter­a­tive-deep­en­ing A* heuris­tic search (IDA*).

The ba­sic idea is this: the ad­van­tage of depth-first search is that you can min­i­mize mem­ory cost by only main­tain­ing the in­for­ma­tion re­lated to the path you are cur­rently try­ing. How­ever, depth-first search can eas­ily get stuck down a fruitless path, while breadth-first search has bet­ter guaran­tees. IDA* bal­ances the two ap­proaches by go­ing depth-first, but giv­ing up when you get too deep, back­ing up, and try­ing a new path. (The A* as­pect is that you define “too deep” in a way which also de­pends on how promis­ing a path seems, based on an op­ti­mistic as­sess­ment.) This way, you simu­late a breadth-first search by a se­ries of depth-first sprints. This lets you fo­cus your at­ten­tion on a small set of ideas at one time.

Once you’ve ex­plored all the paths to a cer­tain level, your tol­er­ance defin­ing “too deep” in­creases, and you start again. You can think of this as be­com­ing in­creas­ingly will­ing to spend a lot of time go­ing down difficult tech­ni­cal paths as you con­firm that eas­ier op­tions don’t ex­ist.

Of course, this isn’t a perfect model of what you should do. But, it seems to me that a note-tak­ing sys­tem should as­pire to sup­port and en­courage some­thing re­sem­bling this. More gen­er­ally, I want to get across the idea of think­ing of your ex­ist­ing re­search method­ol­ogy as an al­gorithm (pos­si­bly a bad one), and try­ing to think about how it could be im­proved. Don’t try to force your­self to use any par­tic­u­lar al­gorithm just be­cause you think you should; but, if you can find ways to nudge your­self to­ward more effec­tive al­gorithms, that’s prob­a­bly a good idea.

In­vent­ing Short­hand/​Symbology

I don’t think writ­ing speed is a big bot­tle­neck to think­ing speed. Even though I “think by writ­ing”, a lot of my time is spent… well… think­ing. How­ever, once I know what I want to write, writ­ing does take time. When in­spira­tion re­ally strikes, I might know more or less what I want to say sev­eral para­graphs ahead of where I’ve ac­tu­ally writ­ten to. At times like that, it seems like ev­ery sec­ond counts—the faster I write, the more ideas I get down, the less I for­get be­fore I get to it.

So, it seems worth putting some effort into writ­ing faster. (Com­puter typ­ing is ob­vi­ously a thing to con­sider here, too.) Short­hand, and spe­cial sym­bols, are some­thing to try.

There’s also the is­sue of space. I know I ad­vo­cate for small cards, which are in­ten­tion­ally limit­ing space. But you don’t want to waste space if you don’t have to. The point is to com­pre­hend as much as pos­si­ble as eas­ily as pos­si­ble. Writ­ing bul­let points and us­ing in­den­ta­tion to make out­lines is an im­prove­ment over tra­di­tional para­graphs be­cause it lets you see more at a glance. Similarly, us­ing ab­bre­vi­a­tions and spe­cial sym­bols will im­prove this.

I’ve tried sev­eral times to learn “proper” short­hand. Maybe I just haven’t tried hard enough, but it seems like ba­si­cally all short­hand sys­tems work by leav­ing out in­for­ma­tion. Once you’re used to them, they’re easy enough to read shortly af­ter you’ve writ­ten them—when you still re­mem­ber more or less what they said. How­ever, they don’t ac­tu­ally con­vey enough in­for­ma­tion to fully re­cover what was writ­ten if you don’t have such a guess. Ba­si­cally, they don’t im­prove read­abil­ity. They com­press things down to the point where they’re hard to de­ci­pher, for the sake of get­ting as much speed as pos­si­ble.

On the other hand, I’ve spent time ex­per­i­ment­ing with changes to my own hand­writ­ing which im­prove speed with­out com­pro­mis­ing read­abil­ity. Pay at­ten­tion to what takes you the most time to write, and think about ways to stream­line that.

Lion Kim­bro em­pha­sizes that you come up with ways to ab­bre­vi­ate things you com­monly re­peat. Ho de­scribes us­ing the Ja­panese sym­bols for days of the week and other com­mon things in his sys­tem. The Bul­let Jour­nal­ing com­mu­nity has cre­ated its own sym­bol­ogy. Per­son­ally, I’ve ex­per­i­mented with a va­ri­ety of differ­ent refer­ence sym­bols which mean differ­ent sorts of things (be­yond the () vs [] dis­tinc­tion I’ve men­tioned).

The Bul­let Jour­nal­ing com­mu­nity has thought a lot about short-and-fast writ­ing for the pur­pose of get­ting things out quickly and leav­ing more space on the page. They also have their own sym­bol­ogy which may be worth tak­ing a look at. (I don’t yet use it, but I may switch to it or some­thing similar even­tu­ally.)

Well, that’s all I want to say for now. I may add to this doc­u­ment in the fu­ture. For now, best of luck de­vel­op­ing ideas!